MSI GeForce 2070 Ventus GP crash.

Joined
Aug 12, 2020
Messages
5
Hello.
I recently acquired a MSI GeForce 2070 Ventus GP 8GB GDDR6. It worked for like few hours or so. After that, it started fail with a driver crash (nvlddmkm stopped working and recover) while playing or using any intensive GPU application. The crash usually happens as soon the game starts, on title screen or, in case of Steam Big Picture, as soon it starts. First freeze for a moment, then crash. System itself doesn't crash.
Computer specs:
CPU: AMD Ryzen 7 2700x 3.7GHz
Motherboard: Gigabyte B450 DS3H (Stock BIOS, F50)
RAM: Kingston Hyper Fury X 2666MHz 8GB x2
GPU: MSI GeForce 2070 Ventus GP 8GB GDDR6
SSD: Sandisk Plus 240GB SATA III
HDD: Seagate Barracuda 2TB 7200rpm
Wifi: TP-Link TL-WN881ND 300mps
PSU: Tempest Gaming 750W
SO: Windows 10 Education

All drivers and system are update to latest version. I found a workaround by changing the GPU power mode to High Performance, but this consumes too much and heat is also higher. However, the GPu doesn't overheat, neither the CPU even on this mode (max temp is 70ºC at max load, on average of 60ºC). On this high performance mode, the crash doesn't happens, however doesn't look like a valid solution.

I tried reinstalling the SO, reinstalling drivers, trying another PSU, trying different drivers versions, change performance mode to High Performance and try the GPu on another computer. Only changing the power mode to high perfomacne works. Also, the GPu worked on an intel based computer (i3, asus motherboard and 16GB RAM9 and didn't fail.

I am not sure what is failing, if it is a incompatibility with AMD, or the PSU is failing (despite I tried a different one, giving the same problem). I was reading it could be the "+12V Rail" on the PSU, and I should get a PSU with a minimum value on this characteristic. But I have no idea what is the minimum for this GPU/CPU.

Anyone have an idea of what could be failing? If that's the PSU, how I can know what PSU minimun I need? It doesn't seems like a GPU or CPU fail and I cannot find that 12V Rail value for this GPU.
 

savormix

LF RTX 3090 Ultra
SERGEANT
Joined
Apr 14, 2020
Messages
50
If you suspect a rail voltage drop, that can easily be checked with HwInfo (you can enable logging if there is a risk of the system crashing rather than just the GPU).
ATX might claim +-5% is fine, but with Turing chips, anything below 11.8 V is suspect and 11.6 V or below is literally asking for trouble.

However, if you do not see a drop below 11.8V on the +12V rail, it is not likely that the rail is the problem in your situation.
 
Joined
Aug 12, 2020
Messages
5
If you suspect a rail voltage drop, that can easily be checked with HwInfo (you can enable logging if there is a risk of the system crashing rather than just the GPU).
ATX might claim +-5% is fine, but with Turing chips, anything below 11.8 V is suspect and 11.6 V or below is literally asking for trouble.

However, if you do not see a drop below 11.8V on the +12V rail, it is not likely that the rail is the problem in your situation.
Thanks for answer. I used HW info before. Using it while a crash occurs causes all data go to 0. But before drop to 0, I've observing the voltaje from PCIe +12V is around 11.9 V, usually, droping to 11.7V, then crash. However, the 8pin input it is always at 1.8~V. But the difference doesn't look too much, so I'm not sure if that is cauisng it.
Captura de pantalla_2020-08-12_17-20-14.png


This is with a replacementPSU. With the PSu I listed before, the normal voltage it is 12.1V and never drops below 12.0.V. So, that's why I am not sure what's the issue.

Could it be a malfunctioning PCIe slot? Damage motherboard? I am more confused by the fact that setting to max performance the GPU doesn't crash at all.
 
Last edited:

flobelix

Well-known member
Global Moderator
Joined
Jan 25, 2007
Messages
39,448
What exact psu model do you have and also list it's specs.
 
Joined
Aug 12, 2020
Messages
5
The current PSU is a Tempest 750W GPSU. It is a Spanish branch. https://www.tempestofficial.com/fue...uente-de-alimentacion-750w-8435443730177.html

Cannot find power specs writed, here is a photo.
Captura de pantalla_2020-08-12_18-37-31.png


Total: 750W
Connectors are:
  • 24Pins cable
  • 2*PATA
  • 5*SATA
  • 2*P6+2
  • 4+4 Pins CPU
I also tried this one:

Also
Specs in english are:
(8-pin): Yes
CPU (8-pin): 1
ATX (24 pines): Yes
PCI Express (6 + 2 pin): 2
PFC): Active
(+12V): 56 A
(+3.3V): 20 A
(+5V): 20 A
(+5Vsb): 2 A
(-12V): 0,5 A

I know maybe being Spanish branches are not very trustable, but are the only ones I can test.
The issue still occurs with all of them.
 

flobelix

Well-known member
Global Moderator
Joined
Jan 25, 2007
Messages
39,448
Well all of these units are very low quality and not even making the advertised 750w. Without a psu to be trusted it is quite hard to say anything. Can you check the card in another pc? Else you can just try to rma it.
 
Joined
Aug 12, 2020
Messages
5
Well all of these units are very low quality and not even making the advertised 750w. Without a psu to be trusted it is quite hard to say anything. Can you check the card in another pc? Else you can just try to rma it.
I tried this card on an Intel I3, with Asus motherboard and 16GB RAM DDR4, which it is not mine. I used that Hiditec PSU I mention before because that PC have only a basic PSU with no 8 pin connector. The card was working pretty fine there. That's why I suspect it is either the Motherboard or PSU, or both. I'm having difficulty to find the failing one, and since is a custom build, I cannot send the entire PC to RMA. I can RMA almost all the computer. Most components are less than 2 months old. That's why I'm trying to figure out which is the faulty one (or ones).
 

savormix

LF RTX 3090 Ultra
SERGEANT
Joined
Apr 14, 2020
Messages
50
Thanks for answer. I used HW info before. Using it while a crash occurs causes all data go to 0. But before drop to 0, I've observing the voltaje from PCIe +12V is around 11.9 V, usually, droping to 11.7V, then crash. However, the 8pin input it is always at 1.8~V. But the difference doesn't look too much, so I'm not sure if that is cauisng it.
View attachment 140292

This is with a replacementPSU. With the PSu I listed before, the normal voltage it is 12.1V and never drops below 12.0.V. So, that's why I am not sure what's the issue.

Could it be a malfunctioning PCIe slot? Damage motherboard? I am more confused by the fact that setting to max performance the GPU doesn't crash at all.
Personally, I would not look at the GPU specific measurements when the GPU is crashing (because when the GPU crashes it is briefly unable to report anything). Even so, we can see a 11.66V reading there - which means the PSU can be at fault here.

The column you should look at (that will maintain its readings) is the "+12V [V]" one. Ideally, it should never drop below 12.0 V (any high tier PSU will manage to do so for its rated wattage). Anything below 11.8 may cause issues for Turing, and anything below 11.6 is almost guaranteed to.
 
Joined
Aug 12, 2020
Messages
5
I appreciated all the help here, but I think I will not find the problem.

I don't know. Everything is very weird and I am not sure what to do. I ordered a new PSU, a Corsair TX750. If this doesn't work, I just RMA RTX2070 and try to fix the issue before buy another card. I am very lost and I cannot found the issue, even at local store where I buy the components doesn't know what is happening, and just recommended me RMA components. The problem however is, they will just give me exact the same hardware, except RTX, which I can get the money back.

Also, I'm finding bottlenecks that causes games to go down 50-45 FPS at 1080p for no reason (they are not top games), when this card shouldn't have any problem with those games. I still find weird that the card works on High Perfomance without crash, but doesn't on Optimized or Adaptable settings.

Anyway, thanks for the help, but I think I'm running in circles and the setup is just fucked up somewhere and I cannot find the issue. A friend gave me a Asus Prime B450M-A Motherboard to try it if the Corsair PSU doesn't fix the issue. But I'm starting thinking it is just a conflict between AMD and Nvidia or just a bad hardware choice by me.
 
Last edited:
You must log in or register to reply.
Top